Note: Descriptions are shown in the official language in which they were submitted.
CA 02771260 2014-03-17
CA 02771260 2012-02-15
WO 2010/021898 PCIYUS2009/053669
METHOD AND APPARATUS FOR DISEASE DIAGNOSIS AND SCREENING USING
EXTREMELY LOW FREQUENCY ELECTROMAGNETIC FIELDS
TECHNICAL FIELD
Disclosed is a novel method and apparatus for diagnosing or screening disease
states in
living organisms by the measurement and analysis of extremely low frequency
electromagnetic
fields. The measurement of such fields is performed at a single point or at
several test points on
or in the body and compared to one or more reference. Information in the time-
varying
electromagnetic field is collected, then processed by diagnostic or screening
algorithms to
provide information about the disease state of the tissue being assessed.
BACKGROUND
It is well accepted that the biological activity of organisms, organ systems
and cells
produces measurable electromagnetic activity. At one end of the spectrum is
the high frequency
activity (alternating current) of neural tissue and at the other end is the
steady state (direct
current) activity hypothesized to indicate abnormal cell or tissue growth. For
example, medical
applications of high frequency (AC) electromagnetic field measurements are
manifest in
electroencephalographic and electrocardiographic devices. More recently,
direct current (DC)
fields have been studied as a method of cancer diagnosis. For example, U.S.
Pat. No. 4,328,809
to B.H. Hirschowitz and U.S. Pat. No. 4,955,383 to M.L. Faupel contemplate
devices and
methods for measuring and analyzing DC electropotentials for disease diagnosis
or screening. In
these inventions, information in the extremely low frequency alternating
current (AC) band is
filtered out through averaging a multiplicity of signals taken over time.
Higher frequency
information is filtered out using active or passive digital or analog filters,
Other manifestations of
this approach have been articulated in U.S. Pat, Nos. 4,407,300 to Davis and
4,557,273 to Stoller
et at. Davis, for example, discloses the diagnosis of cancer by measuring the
electromotive forces
generated between two electrodes applied to a subject.
If measurements are taken from several test points on the body, as
contemplated in the
aforementioned Hirschowitz and Faupel patents, as well as in U.S. Pat. Nos.
4,416,288 to
Freeman and 4,486,835 to Bai, comparisons of the averaged DC potentials from
the plurality of
CA 02771260 2012-02-15
WO 2010/021898 PCT/US2009/053669
test points may be of particular interest. Furthermore, the averaged DC
voltages may be further
analyzed by discriminant function analysis, as disclosed in particular by the
aforementioned
Faupel patent.
These disease diagnosis techniques using only DC electropotentials sacrifice
information
(e.g., low frequency AC information) for ease of processing afforded by a
singular filtered and/or
averaged measurement from each test site on the body. Unfortunately, this loss
of information
may compromise diagnostic accuracy of many disease states. For example, public
disclosure of
clinical studies involving analysis of DC potentials indicate that while there
may be some degree
of diagnostic accuracy for large (palpable) cancers, the same approach appears
to be relatively
ineffective for small (nonpalpable) cancers. Since it is well recognized that
early detection of
disease states offers the best chance for patient survival, improvement in
this capacity over
previous disclosures is indicated. Moreover, previous pattern recognition
techniques used for
analysis of electropotential fields may be overly simplistic in that they do
not take into account
the complexities of biological systems and their disease states. To this
point, it is known that the
biology and concomitant electromagnetic activity of malignant tumors, for
example, change over
time. In order to maximize effectiveness, novel diagnostic and screening
techniques based on the
measurement of electromagnetic fields must take into account both the short
term changes in
electrical activity (e.g., extremely low frequency AC fields) as well as the
longer term changes
which occur as a disease state progresses. Failure to do so results in major
deficiencies leading to
diagnostic inaccuracy.
For example, changes in extremely low frequency alternating current (ELFAC)
may
differ for malignant vs. benign tumors, because the gating mechanisms
controlling ion transport
across the epithelial tissue layer can differ between the diseased and
nondiseased condition, or
other reasons. It is known that malignant epithelial cells lose, to varying
degrees, the ability to
transport ions and fluids across the epithelial layer. It is this low
frequency time-varying
phenomena which is lost by restricting analysis of electrical signals to an
averaged and/or filtered
DC component.
In addition, changes over the longer term indicate that the electromagnetic
behavior of
small malignant tumors may be very different from that of larger tumors, which
have been found
to produce redox potentials as a result of the degradation of tissue within
the core of the tumor.
2
CA 02771260 2012-02-15
WO 2010/021898 PCT/US2009/053669
Another factor is smaller tumors may be more metabolically active and
therefore more relatively
depolarized than larger tumors.
It follows then that analysis of important electromagnetic changes must take
these factors
into account in order to maximize diagnosis and screening of disease states.
SUMMARY
Disclosed is a novel and improved method and apparatus for integrating
electromagnetic
information over time in the form of a set of novel pattern recognition
outputs for the diagnosis
and screening of disease states. The important diagnostic inputs to pattern
recognition are
referred to here as extremely low frequency alternating currents (ELFACs).
Such method and
apparatus operate to measure and analyze electromagnetic activity from regions
of diseased
tissue on or internal to a living organism.
Further disclosed is a novel and improved method and apparatus for pattern
recognition
which takes into account the biological changes which occur as a disease state
progresses. This is
accomplished by integrating non-electromagnetic information (e.g., size or
area of diseased
tissue as indicated by an imaging study or palpation of a tumor) with the
electromagnetic
information.
Further disclosed is a novel means for identifying and discriminating
diagnostically
important ELFAC activity from electromagnetic activity due to noise at the
interface between the
organic tissue and the measurement apparatus. This is accomplished by constant
measurement of
ELFAC activity from the point a measureable signal is achieved from the
organism but recording
and analyzing only the portion of the signal indicating that ELFAC activity
has achieved a noise-
minimized and therefore diagnostically useful state. Since this is expected to
vary on a subject by
subject basis, a novel means of determining noise minimized performance using
pattern
recognition is disclosed by the present invention.
In some embodiments, the invention provides a novel and improved method and
apparatus for screening an organ system of a living organism for a disease
condition. In some
embodiments, noise-minimized ELFACs are recorded from several locations
involving an organ
system such as the prostate gland or breast. The resultant information is
analyzed using nonlinear
pattern recognition techniques to determine if a disease state exists in an
organ system for which
no other disease related symptoms are observable. Improvement in screening
accuracy may be
3
CA 02771260 2012-02-15
WO 2010/021898 PCT/US2009/053669
achieved by integrating subject variables such as chronologic age or organ
characteristic (e.g.,
size) information along with ELFAC data into the screening pattern recognition
program.
Still further embodiments of the present invention provide a novel and
improved method
for ELFAC diagnosis of malignant disease states. In this embodiment,
recordings of noise-
minimized ELFACs from locations on and near tissue suspected of harboring
malignant activity
are analyzed using specific nonlinear pattern recognition programs which
incorporate ELFAC
data with known symptomatology of the suspect tissue, such as tumor size,
level of suspicion
from imaging studies, age of subject, etc. The goal of this approach is to
take advantage of the
interaction between progressive biological changes which occur during
carcinogenesis, tumor
growth, metastasis and their resultant electromagnetic alterations. In
situations where bilaterality
of the organ system is accessible by the measurement apparatus, as with the
breast or extremities,
comparison of the suspect breast or extremity with the unaffected or opposite
breast or extremity
can be used to provide a set of internal control measures. Likewise, a control
or reference point
can be an external solution of normal saline with any voltage offset
calibrated out of the
measurement system.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 is a block drawing of the apparatus of some embodiments of the present
invention;
FIGs. 2a and 2b are each a sectional diagram of an electrode for the apparatus
of FIG, 1;
FIG. 3 is a flow diagram of the measurement operation of the apparatus of FIG,
1 used to
obtain noise-minimized ELFAC signals;
FIG. 4 is a flow diagram representing inputs and outputs of the disease
decision pattern
recognition program.
FIG. 5 is a diagram showing the differentiation between spurious
electromagnetic activity
due to skin/electrode equilibration and noise-minimized ELFAC signals.
FIG. 6 is a schematic for a passive two-pole band pass filter employed by some
embodiments of the invention.
FIG. 7 is a schematic for an active two pole band pass filter employed by some
embodiments of the invention.
FIG. 8 is a graph depicting the frequency response of a bandpass filter
employed in some
embodiments of the invention.
4
CA 02771260 2012-02-15
WO 2010/021898 PCT/US2009/053669
DETAILED DESCRIPTION
The block diagram for the apparatus in accordance with some embodiments of the
present invention is disclosed in FIG 1. The apparatus performs the functions
necessary for
obtaining and analyzing noise-minimized ELFAC data and integrating those data
with other
information to produce a disease diagnosis. For the purposes of illustration,
the apparatus 10 will
be discussed as configured for the diagnosis of skin cancer, although it
should be recognized that
the method and apparatus can be reconfigured and similarly employed for
screening or
diagnosing other portions or organs of a living human or animal, such as the
mammary gland,
prostate gland, colon, lung, naseo-pharynx, or other organ systems.
In FIG. 1, a human subject's skin surface 12 may have a suspicious lesion 14
visible on
the right forearm. Clinical inspection of the lesion may be equivocal and a
more certain diagnosis
may be needed. In this case, the location of the lesion is known and a
multiple sensor array 16 is
applied to the area of suspicion and a reference sensor is applied to the
mirror image positions on
the opposite (left) forearm (not shown). It should be recognized that the
opposite forearm is
being used as the reference. In other examples, the reference could be a
saline solution or other
external reference standard, or an undiseased portion of the same organ, or
other undiseased
tissue of the patient, etc. Assume, however, for the purposes of discussion,
that the area of a
lesion 14 is not known, as in the case of screening for breast cancer. In this
situation, a large
array of electrodes 16 would be used to identify any area of suspicion
occurring on either breast,
as indicated by asymmetries in the ELFAC activities between the two breasts.
This illustrates
that different embodiments of the device and method of the present invention
contemplate the
use of a variety of different electrode arrays depending on the specific
application, and whether
the application is for disease screening or diagnosis. Likewise, the number of
recording sensors
16b may differ depending on the specific application. However, in both
examples significant
disease can be detected using pattern recognition of ELFAC data in conjunction
with other
clinical information. Once accessed, analog electromagnetic signals are passed
through a
bandpass filter 18, converted to digital form at analog to digital converter
20, processed at CPU
22 using ROM 24 and RAM 26, and finally displayed at 28. During processing,
the measured
value(s) are compared to the reference value(s), and adjusted for other
characteristics if desired.
The result is then compared, in the case of screening activity, to the
reference to suggest the
likelihood of disease, and in the case of diagnosis to known patterns to
suggest the possibility
CA 02771260 2012-02-15
WO 2010/021898 PCT/US2009/053669
and type of disease. The actual steps describing the invention are
subsequently disclosed more
specifically in connection with FIGS. 2, 3 and 4.
The ELFAC sensing electrodes can be applied either individually or as a set of
sensors on
an adhesive flexible backing , depending on the application. In both cases,
effective spacing of
electrodes should be maintained, so that overlap in measurements is minimized.
For example, the
distance between individual sensors 16b should be at least two times the
diameter of the sensor
area in contact with the skin or other tissue. If sensors 16b are applied
individually, the
technician must be trained to position the sensors using the appropriate
distances. Ideally, the
electrodes 16b should be of a type which do not cause a substantial battery
effect due to a
dissimilar metals reaction. Most modern electrodes have a solid rather than
liquid gel, but the
solid gels don't penetrate the stratum corneum, so the signal to noise is not
optimal. Another
approach is a sensing electrode that consists of a silver or other conductive
component 30 having
an electrical lead 32 connected to the measurement apparatus 10. This
electrical connection is
secured to the silver component 30 by a conductive top piece 34. A thin layer
of silver chloride
36 is deposited on the surface of the silver component 30. This system is
embedded in
nonconductive plastic or plastic backing sheets 38. The interface between the
organ tissue 12 and
the silver chloride surface 36 is mediated by a semi-liquid electrode cream,
paste or gel 40 of a
known type such as Synapse electrode creme manufactured by Rose Labs, Inc. The
primary goal
of the electrode cream 40 is to provide a conductive pathway through the
electrically resistive
corneal layer of the skin 12. Alternatively, the present invention
contemplates that the corneal
layer of the skin 12 can also be breeched mechanically, either by using
recording mini-or micro-
electrodes of known types (e.g., Beckman Coulter, Inc. Fullerton CA)or by
using an array of
needle electrodes on a backing sheet which penetrates just beyond the corneal
layer of the skin
12 but are of such small diameter not to cause tissue damage. This alternative
electrode design is
shown in FIG. 2B at 42 and does not require the use of an electroconductive
cream, paste or gel
40. In this embodiment, the conductive component 30 contains numerous small
penetrating silver
or platinum electrodes 42, each of these electrodes could be connected to its
own lead which
connects to the measurement apparatus 10 or summed to one lead (not shown)
connected to
measurement apparatus 10.
Alternatively, the stratum corneum, which is responsible for more than 90% of
the
electrical resistance of the skin, can be removed by controlled use of laser
light or a heated wire.
6
CA 02771260 2012-02-15
WO 2010/021898 PCT/US2009/053669
In this mode, a laser light in the near infrared range of the electromagnetic
spectrum can be
concentrated on a thin layer of a dye which absorbs the light energy and
dissipates the stratum
corneum without penetrating to the dermis below. One or more pores of varying
size can be
produced in this or any other suitable manner. After the stratum comeum has
been selectively
compromised in this manner, the electrode is placed over the site and ELFAC
potentials
measured. In this instance, the electrode 16 or electrodes can be placed
directly in the pore or
pores thus created, without the need for the electrode cream or gel 40, and
without the need for a
needle electrode 42.
The measurement device 10 contains multiple recording inputs having electrode
leads 32
which are affixed to the subject 12. A reference lead is affixed to the
patient in a nondiseased
area or alternatively to an external solution of normal saline or other
reference standard.
Although the example demonstrates the use of a separate reference electrode,
each recording
electrode could double as a reference electrode for the other electrodes. In
this embodiment, each
recording electrode except one is scanned relative to the remaining electrode.
Next, a different
"recording" electrode is automatically selected to be the reference electrode,
and so on until each
recording electrode has functioned as a reference electrode as well. Since
this would result in
multiple measurements for each recording electrode, the arithmetic average,
median or mode
could be taken as representative of each recording channel.
The measurement apparatus 10 can utilize a selective bandpass filter 18,
allowing, for
example, the analysis of ELFAC signals in the frequency range from 0.01 Hz to
0.1 Hz.
Although the present example for cancer diagnosis may utilize signals within
this particular
frequency range, the present invention also contemplates using bandpass
filters sensitive to other
low frequency ranges, depending on the type of disease diagnosed or screened
for In general,
higher ranges may apply to cases where there is higher electrical resistance
due to skin for
example that must be bypassed to detect breast cancer. Cancers that occur on
the linings inside
the body such as esophageal lung or cervical don't have this issue and may be
closer to true DC
(i.e., lower frequency. Until the studies are performed we won't have actual
numbers.. The
bandpass filter 18 may include one or more filters of a known design, as
described in the
methods for this.
7
CA 02771260 2012-02-15
WO 2010/021898 PCT/US2009/053669
A band pass filter can in principle be constructed by combining a low and high
pass filter
in cascade. Figure 6 shows such a filter. The first part (CIRO will pass high
frequency signals
while the second part (C2R2) will pass the low frequencies (or rejects high
frequency signals).
However, the filter cannot be considered as a simple cascading of a high and
low pass filter since
the second part loads the first part. As a result, the overall transfer
function is not simply the
product of the individual transfer functions of the high and low pass
sections.
The above examples of filters are called passive filters since they do not
make use of
amplifiers. Among the disadvantages of such a filter is that there is no gain
and that the load
resistor RL influences the transfer characteristic. A better way to build
filters for low to mid-
frequency applications is to use operational amplifiers. Such filters are
called active filters. The
advantage of active filters is that one can provide amplification, and have a
filter whose
characteristic is independent of the load.
A simple active band pass filter (two pole system) is shown in Fig. 7. The
frequency
response (Bode graph) of the gain is given in Fig. 8.
The band pass filter separately filter the signals on each of the input leads
32 which then
pass each of the filtered signals via a separate channel to a multiple input
analog-to-digital
converter 20. Alternatively, bandpass filtering could occur in the digital
domain post analog-to-
digital conversion. In addition. the bandpass filter 18 could constitute an
individual filtering
mechanism for each channel, providing filtering only for that channel with
each filtered output
connected to the input of the analog-to-digital converter 20.
The analog-to-digital converter 20 ideally should be capable of multiple input
multiplexing such as that manufactured by National Semiconductor, Inc. and
designated as
ADC808. For very large measurement arrays, such as those contemplated for
breast cancer
screening, more than one multiple input analog-to-digital converter may be
necessary, the precise
number of converters determined by the channel converters capacity and the
number of channels
required for a specific application.
The analog-to-digital converter 20 converts each channel's analog signal to a
digital
signal which is relayed via a separate output channel to the multiple inputs
of a central
processing unit 22. The central processing unit is a component of the larger
control system which
also contains RAM 24 and and ROM 26. A stored program in the central
processing unit 22
8
CA 02771260 2012-02-15
WO 2010/021898 PCT/US2009/053669
controls signal acquisition and sampling rate, and then processes the digital
input data to produce
an output to the user regarding the disease state of the tested tissue. Other
relevant data such as
patient age or size of lesion can be inputted by using a standard computer
keypad or touch
sensitive screen of conventional type or other input device or method. The
central processing
unit then intregrates this information with the ELFAC data using preprogrammed
pattern
recognition algorithms. The final output to the user is then fed to a display
device 28 such as a
computer monitor or printer. The output could also be stored locally or
remotely in a network or
other memory system. The output can constitute a numerical answer as to the
probability of the
disease state existing, a yes/no answer as to whether the disease in question
exists, or a scalar
result indicating the severity of the disease, and/or a false color image,
depending on the specific
application.
The function and operation of the ELFAC device will be understood from two
examples
of the steps which embody the basic methodology. The first example summarizes
the method for
disease screening while the second example characterizes the invention using a
diagnostic
format.
In the case of screening, the position and disposition of a lesion cannot be
identified
because the subject is asymptomatic. In this case, a relatively large array of
electrodes 16is
positioned either on the surface 12 of the site in question. If the suspected
site can be assessed
externally through the skin, then the electrodes may be placed on the skin. If
more invasive
procedures are indicated, the electrodes can be placed internally at the
subject organ or site. In
the case of breast cancer screening, preferably, the entire surfaces of both
breasts would be
measured, since the user is unaware if and where a malignancy exists. Once the
electrode array
16 is positioned, the reference electrode is placed on an undiseased area of
tissue or in a
reference standard. Then the ELFAC activity between the reference electrode
and each of the
measurement electrodes 16 and is immediately measured, bandpass filtered, and
processed by a
preprogrammed algorithm which determines if and when diagnostically useful
ELFAC readings
have been obtained from a given test subject. At this point, each individual
voltage reading in the
waveform is preserved for pattern recognition. This differs from previous
techniques in which
the apparatus filtered even low frequency AC information by a combination of
active filtering
and arithmetic averaging of multiple signals taken over time. The approach
taken by previous
manifestations was to identify only a representative direct current (DC)
component by utilizing
9
CA 02771260 2012-02-15
WO 2010/021898 PCT/US2009/053669
this selective low pass filtering and signal averaging. Thus any information
in the ELFAC
component was lost to analysis.
In the case of diagnostic analysis, a sensor 16 is placed at a suspicious
site, such as a
lesion. The number of electrodes and their positioning will depend, in part,
on the size of the
suspicious site. Generally, the suspicious site and some surrounding tissue
should be analyzed.
As with the screening process, a reference electrode is placed at either a
reference cite or in a
reference standard. A reference site could be a corresponding position on a
mirror site (e.g. left
arm vs. right arm) or an undiseased portion of the same organ or tissue, or
other undiseased
tissue, ELFAC activity between the reference electrode and each of the
measurement electrodes
16 and treated similarly as described above for screening.
In order to determine whether and when noise is minimized and therefore
diagnostically.
useful ELFAC signals are obtained from a given subject, monitoring of these
signals must be
continuous from the time the last electrode is placed on the subject until
representative ELFAC
signals are obtained. It is well recognized that noise is produced at the
skin/electrode interface
due to the electrically resistive characteristics of the skin's stratum
corneum. For skin surface
electrodes to be effective transducers of electromagnetic fields generated by
subsurface organ
systems, the corneal barrier must be breeched. However, using even the most
highly conductive
electrode pastes and gels, achieving signal equilibration takes time; several
minutes or even
longer for a given subject. But more problematic is the fact that
equilibration time can vary
substantially on an individual basis and can even vary within an individual at
different times
and/or test site locations. This is because skin resistance and other factors
related to transdermal
activity vary on an individual basis. Previously the timing and duration of
electromagnetic field
measurement was predetermined and the same for each subject essentially built
into the machine.
This could lead to diagnostic errors because some subjects could be measured
before
equilibration was complete. hi these cases, it was noise rather than signal
which was measured
and analyzed. The present invention circumvents this problem by continuous
measurement and
monitoring of ELFAC activity. This is accomplished by analyzing each incoming
signal for
stability on an individual basis. For example, the typical manifestation of
uncompleted signal
equilibration is a progressively decreasing (but not increasing) electrical
potential seen soon after
a measurement electrode is applied and the corneal layer is in the process of
being penetrated by
the electroconductive paste or gel. Eventually, once equilibration is
complete, the informative
CA 02771260 2012-02-15
WO 2010/021898 PCT/US2009/053669
signal of slowly rising and falling ELFAC potentials may be observed. In some
embodiments,
the present invention will not start recording and analyzing electromagnetic
signals until the
pattern recognition software in central processing unit 22 identifies the
characteristic ELFAC
waveforms which indicate that equilibration has taken place. This means that
recording time and
duration are not preset and the same for each subject, but that recording and
analysis of
electromagnetic activity will not take place until signal-to-noise is
maximized for each subject
tested. Each patient and each procedure therefore is assessed individually and
independently to
achieve noise-minimized data in each test in each patient. In some
embodiments, the signals
obtained during equilibration may be recorded, but not used in the screening
or diagnostic
analysis.
As described above, once noise-minimized ELFAC signals are identified, they
are
recorded and held in RAM memory 26 for analysis. The preprogrammed analysis
software then
employs nonlinear pattern recognition techniques of known types such as
artificial neural
networks or classification decision trees (e.g., CARD. Specifically, all
recorded voltages within
the predetermined frequency range are fed into the pattern recognition program
along with
optional subject variables such as patient age, lesion size, family history of
breast cancer (or
disease being screened for, results of imaging studies, etc. The present
invention contemplates
that different pattern recognition programs will be employed for subsets of
subjects based on
these and other key variables. For example, the biological activity of
malignant breast tumors
may be different in premenopausal women as opposed to postmenopausal women
because of
differences in the hormonal milieu in some embodiments, these differences can
be taken into
account in the analysis. Small breast cancers which are nonpalpable may be
more metabolically
active than larger tumors, whose central core often becomes necrotic.
Likewise, metabolic
activity of skin lesions may be tied to physical appearance, with those
lesions appearing as
darker, raised mole-like structures more likely to be highly metabolically
active.
The flow diagram in FIG. 3 provides an example of inputs, central processing,
and the
output of a pattern recognition program for the diagnosis of a skin lesion. A
start switch 44
begins operation of the central processing unit 22, initializing processing
operations 46. The
initializing process brings the various components of the device 10 into
operational mode,
including resetting and activation of control registers to read data 48 from
the analog-to-digital
converters 20. As opposed to prior art devices, the current invention does not
initiate a
11
CA 02771260 2012-02-15
WO 2010/021898 PCT/US2009/053669
predetermined multiple measurement period at 48. Instead, data is read
continuously until
equilibration is completed at 50 and diagnostically useful ELFACs for each
recording channel is
identified. The continuous assessment of data at 50 is recycled back through
step 48 until all or
most of the channels are determined to carry noise-minimized ELFAC signals.
FIG. 5 represents electromagnetic data as a function of time and amplitude.
ELFAC
signals digitized at 48 are determined at 50 to be either obscured by noise
112 or to have reached
a noise minimized state 114. Only signals which have reached a noise minimized
state 114 are
recorded (saved) at 52 and stored for processing at 54.
Should a significant number of channels be determined as unable to transmit
noise-
minimized ELFAC signals within a reasonable amount of time (e.g.,
approximately 15 minutes)
the operation would shut down and the central processing unit 22 would
transmit instructions via
output display device 28 for the medical technician to check contact points or
employ other
trouble shooting methods. In the case that some channels are unable to
transmit noise-minimize
ELFAC signals, the channels that are able to transmit such signals will be
used, provided that the
number of non-trasnmitting channels is not significant. That is to say that if
substantially all the
channels are transmitting, the transmitting channels will be used. By
"substantially all" it is
meant that at least 90% of all channels are able to transmit noise-minimized
ELFAC signals.
Once noise-minimized ELFAC signals have been identified, they are captured and
stored
in memory 54. This is in contrast to prior art devices in which only the
averaged DC component
is stored in memory for processing. The pattern recognition module 58
incorporates nonlinear
pattern recognition programs of known type such as artificial neural networks
or decision trees
which integrate ELFAC data and patient variables 56 such as patient age,
lesion size or shape,
level of suspicion from imaging studies, etc. to obtain a pattern recognition
result 60 which leads
to a probability statement 62 of whether a suspicious lesion is malignant (as
used in a diagnostic
mode) or whether an organ system may harbor an occult malignancy (as used in a
screening
mode). The resultant probability statement can be used to indicate CANCER at
64 if it exceeds a
certain value, such as 0.05 (5%) or NO CANCER if it does not exceed this
value, as in FIG. 4 at
66. This indication, or the underlying probability or other interpretation,
can be output to a
display device such as a monitor or printer and/or stored in memory. If the
present invention is
utilized in diagnostic mode, the probability cut-off point for indicating
cancer can be reduced to
minimize false negative results (i.e., missed cancers) in symptomatic
populations with a high a
12
CA 02771260 2012-02-15
WO 2010/021898 PCT/US2009/053669
priori prevalence of disease. If, on the other hand, the present invention is
utilized in a screening
setting, in which the a priori prevalence of disease is relatively low, the
process at 76 could be
calibrated to indicate cancer at a higher probability cut-off point. Depending
on the specific
application, the output to the user at 64 or 66 could, in addition to
indicating CANCER or NO
CANCER, also indicate the probability or likelihood that the diagnosis is
correct, or a scalar
output indicating disease severity. Once the output 64 or 66 is produced, the
program terminates
at 68.
An example of a configuration of a pattern recognition module at 58 is given
in FIG. 4.
The pattern recognition program initiates at 70, Subject clinical information,
inputted via
standard devices at 56 such as a keypad or from a menu on a touch sensitive
screen or any
suitable input device, constitute the initial steps in the sequence. For
example, at 72 the program
diverges depending on whether the subject is pre-or post menopausal, if the
subject is
premenopausal, then the numerical value corresponding to day in menstrual
cycle is entered at
74. The program diverges again at 76 if the subject is in a hormonally active
segment of the
menstrual cycle. This is because disease states are known to be influenced by
systemic hormonal
changes. If the subject is in a hormonally active segment, then the pattern
recognition program
might weight more heavily differences in point to point ELFAC potentials
between the two
mirror-image organ systems or locations 78, such as between two extremities or
between two
breasts. If significant differences are found at 78, then the PRR value 60
would indicate a higher
probability of cancer, while if these differences are not observed then
waveform characteristics
80 such as frequency, electrical potential at peak and/or trough of the ELFAC
wave are
examined to determine the probability outcome at 82 or 84. If on the other
hand the patient is not
in a hormonally active segment of the menstrual cycle, then maximum
differences from electrode
sites near the area of the symptoms (e.g., on the same breast or extremity)
may be more heavily
weighted by the pattern recognition program at 86. The program may then
proceed to the output
probability at 88 if a positive response is achieved, or proceed to 80 for
waveform evaluation and
subsequent probability output. The pattern recognition values may differ
depending on whether
the input comes from 78 or 86.
Moving now back to 72 in FIG. 4, if the patient is not premenopausal, then it
is irrelevant
to enter a value corresponding to day in menstrual cycle, as at 74. For the
purposes of
illustration, the data flow could proceed on the basis of whether the
suspicious lesion being
13
CA 02771260 2012-02-15
WO 2010/021898 PCT/US2009/053669
tested is palpable or not at 92. If more precise size information from imaging
studies is available,
these data could also be used. If the lesion is nonpalpable, then the program
may proceed to
determine if the ELFAC waves indicate general depolarization in the area of
the suspicious
lesion (relative to measurements taken from the opposite breast or extremity)
at 94 with eventual
probability outputs at 96, or 104 or 106 via 98. If on the other hand the
lesion is palpable at 92,
then the pattern recognition program at 100 would weight maximum ELFAC
differences from
electrode sites near the area of the symptoms (e.g., on the same breast or
extremity) rather than
weighting overall depolarization relative to the opposite breast or extremity.
If ELFAC differences are above a preset threshold as determined by the
preprogrammed
algorithmat 98, then a higher probability of disease is indicated at 106. In
each case at 94 and
100, negative outcomes would lead to waveform characteristic analysis at 98
with negative
outcomes leading to statements of lower probability of disease at 104 and
positive outcomes
leading to statements of higher probability of disease at 106.
Although FIG. 4 provides one embodiment of a pattern recognition flowchart,
the present
invention also contemplates that other biological variables could be added to
the pattern
recognition module 58. For example, lesion palpability 92 could be integrated
into the decision
sequence for premenopausal subjects as well. In addition, the probability
statements at 80, 84,
86, 90, 96, 102, 104, and 106 could be integrated with level of suspicion
indexes from other
diagnostic studies or previous ELFAC tests. The number and types of variables
would depend on
the specific application of disease diagnosis or screening. For example, a
disease screening
pattern recognition module would not include lesion characteristics such as
palpability or size 92
in FIG. 4. Instead, there would be greater reliance placed on demographic
variables such as
subject age or family history of the disease state being screened.
Thus, different decision cut-off points or submodules (e.g., neural nets,
decision trees)
such as 78, 82, 88, 94, 98, and 100 in FIG. 4 are required for different
biological situations in
order maximize the effectiveness of utilizing electromagnetic fields for
disease diagnosis and
screening.
The method and apparatus of the present invention is designed to provide
greater
accuracy in diagnosing and screening for disease states using biologically
tuned pattern
recognition of extremely low frequency alternating current (ELFAC) signals.
The signals are
measured from a number of different sites on the body involving a known or
suspicious disease
14
CA 02771260 2012-02-15
WO 2010/021898
PCT/US2009/053669
site. Comparison of the informative aspects of the signals from the different
sites such as
periodicity, peaks, troughs, slopes and other information are integrated with
subject variables to
provide a greater accuracy in detecting and diagnosing disease states such as
epithelial
malignancies.
